An annual international competition to find the best Machine Translation systems has announced the 2009 winners. The Department's team, was ranked the top Arabic-to-English translation system. The team was lead by Dr Bill Byrne and Dr Adrià de Gispert and including Graeme Blackwood and Jamie Brunning from the Department and Gonzalo Iglesias from the University of Vigo, Spain. The system was placed first in both the Single System Track and the System Combination Track.
The competition, run by the US National Institute of Standards (NIST), is part of an ongoing series of evaluations of human language translation technology. NIST defines a set of translation tasks and a set of common resources so that systems can be directly compared to each other. The Cambridge system was based on a hierarchical phrase-based decoder implemented with weighted finite state transducers. The system also performed multiple morphological analyses of Arabic text to improve translation of `noisy' text, such as web pages and blogs. The competition is a showcase for postgraduate research in the Department and is closely watched by industry and academic researchers worldwide.